Patent Grant
8130535
The disclosure relates generally to testing technologies for electronic circuits and, more specifically, to a method and apparatus for testing magneto-resistance random access memory.
Magneto-resistance random access memory (MRAM) is attractive as a form of memory in many applications. It is capable of being integrated on-chip within the standard integrated circuit (IC) architecture. MRAM is non-volatile, not requiring power to maintain a memory state, i.e., the memory state is not lost in the event of a power loss. MRAM is based on writing and reading a tunneling resistance value of a magneto-resistive tunnel junction (MTJ). A MRAM bitcell includes an MTJ cell and a switch, which may be a transistor, for example. An MTJ cell is a junction having a fixed polarization magnetic layer and a free (i.e., re-writable) polarization magnetic layer separated by an electrically insulating tunneling layer.
In conventional MRAM, the free layer polarization is written by imposing an external magnetic field of sufficient strength. The tunneling resistance may commonly have two values depending on whether the free layer polarization is parallel or anti-parallel to the fixed layer. The tunneling resistance is lower in the parallel polarization state than in the anti-parallel polarization state.
Spin torque transfer (STT) MRAM is a form of MRAM where a reversible external magnetic field is not required to toggle the free layer polarization. This is accomplished by applying a sufficient density of spin polarized electron current through the junction. Parallel polarization occurs when the electrons in a write current flow from the fixed to the free layer. Anti-parallel polarization occurs when the write current flows in the opposite direction. Read current densities are generated by voltages that are too small to switch the free layer polarization, but otherwise operate in a similar manner, to measure the tunneling resistance and thereby determine the state of the MTJ cell.
Just as in standard IC testing, large arrays of memory may require testing of every memory cell. One part of testing is the characterization of read/write (R/W) speeds and critical currents, current densities and voltages. The amount of current density (as determined by the amount of applied voltage) required and the pulse width required to perform either read or write functions reliably may be measured.
STT MRAM has the capability to provide relatively high speed, low power, non-volatile memory. The (R/W) speed may be less than 1 nanosecond (ns). Pulse width characterization should therefore extend from less than 1 ns to effective DC. However, conventional test equipment may not be capable of providing sub-nanosecond pulse width signals. Standard testers, such as the Advantest T5585 memory tester (Advantest America Corporation, Santa Clara, Calif.) may have a minimum pulse width of 1 ns, for example, and may also have a maximum available pulse width (e.g., commonly a few ms). Furthermore, off-chip and on-chip capacitance, inductance and resistance existing between the tester port and the MTJ cell on-chip may prevent accurate measurements at the highest speeds, i.e., sub-nanosecond pulse widths, preventing an accurate determination of how fast the MTJ cell may be capable of operating in read and write modes.
In the interest of performance, including without limitation, power consumption, where battery life may be extended in portable applications, it is desirable to determine the least amount of power required to read and write to a memory bitcell based on the minimum pulse width and voltage required.
In one embodiment, a method for generating a signal having a variable pulse width on a semiconductor device, includes receiving a first clock signal on the semiconductor device; and receiving a second clock signal on the semiconductor device having a variable delay relative to the first clock signal. The method also includes outputting a final signal having a rising edge triggered by the first clock signal and a falling edge triggered by the second clock signal. The method further includes providing the final output signal to circuitry on the semiconductor device.
In another aspect, an on-chip variable pulse width signal generating circuit includes a first on-chip latch configured to receive a first clock signal; and a second on-chip latch configured to receive a second clock signal delayed from the first clock signal by a variable amount. The circuit also includes an on-chip set-reset latch configured to receive a signal output from the first latch and a signal output from the second latch, and further configured to output a pulse of time duration based on a delay of the signal from the second latch relative to the signal from the first latch.
In yet another aspect, an on-chip variable pulse width signal generating circuit for testing a spin-torque-transfer (STT) magnetic tunnel junction (MTJ) memory cell includes a first on-chip latch configured to receive a first clock signal. The circuit also includes a second on-chip latch, and an on-chip set-reset latch. The second on-chip latch is configured to receive a second clock signal delayed from the first clock signal by a variable amount. The on-chip set-reset latch is configured to receive a signal output from the first latch and a signal corresponding to an output from the second latch, and further configured to output a pulse of time duration based on the delay of the second clock signal relative to the signal from the first clock signal. The STT-MTJ memory cell is configured to receive a final signal based on the output pulse from the set-reset latch. The STT-MTJ memory cell includes a bit line, an MTJ, a transistor source, and a transistor gate. The bit line is adapted to receive a bit line voltage. The MTJ has a free magnetic layer coupled to the bit line and a fixed magnetic layer coupled to a drain of a transistor. The source of the transistor is coupled to a source line adapted to receive a source line voltage. The gate of the transistor is adapted to receive a signal corresponding to the output pulse of the set-reset latch.
In still another aspect, an on-chip variable pulse width signal generating circuit including a first on-chip means for receiving a first clock signal; and a second on-chip means for receiving a second clock signal delayed from the first clock signal by a variable amount. The circuit also includes a third on-chip means for receiving a signal output from the first receiving means and a signal output from the second receiving means, and for outputting a pulse of time duration based on a delay of the signal received by the second receiving means relative to the signal received by the first receiving means.
The foregoing has outlined rather broadly the features and technical advantages of the present disclosure in order that the detailed description of embodiments of the disclosure that follows may be better understood. Additional features and advantages of the disclosure will be described hereinafter which form the subject of the claims of the disclosure. It should be appreciated by those skilled in the art that the conception and specific embodiments disclosed may be readily utilized as a basis for modifying or designing other structures for carrying out the same purposes of the present disclosure. It should also be realized by those skilled in the art that such equivalent constructions do not depart from the spirit and scope of the disclosure as set forth in the appended claims. The novel features which are believed to be characteristic of the disclosure, both as to its organization and method of operation, together with further objects and advantages will be better understood from the following description when considered in connection with the accompanying figures. It is to be expressly understood, however, that each of the figures is provided for the purpose of illustration and description only and is not intended as a definition of the limits of the present disclosure.
For a more complete understanding of the present disclosure, reference is now made to the following descriptions taken in conjunction with the accompanying drawings.
A method and apparatus are disclosed for testing memory on an integrated circuit chip. The chip includes memory bitcells and on-chip circuitry adapted to provide a variable pulse width signal to each memory bitcell based on one or more external clock signals received by the chip.
In
The integrated circuit devices 125A, 125B, and 125C may be coupled to a microprocessor or other microelectronic device. The integrated circuit devices 125A, 125B, and 125C may be packaged with the microprocessor and further incorporated into a communications device. For example, the circuit may be embedded in a mobile phone or a communications base station.
In this configuration, as shown in
A larger difference in the resistance of the two states makes a determination of the memory state of the STT MRAM bitcell 200 easier and more reliable during read. During write, at least two factors affect the ability to properly write a reliable 0-state or 1-state to the MTJ 201. First, the voltage applied between the bit line BL and source line SL should be sufficient to produce a current through the MTJ 201 that at least exceeds a critical current when the word line signal, WL, is sufficient to open the gate G of the transistor 202. Second, the word line signal, WL, should be applied for a sufficient period of time for the polarized electron current to write the desired state 0 or 1 to the MTJ 201. Assuming the critical current is known, and the voltage between the bit line BL and the source line SL and word line WL voltage are appropriately chosen to provide a sufficient current, it is the time period, i.e., the “pulse width” of the word line signal, WL, that controls producing the desired polarization states.
In a read mode, i.e., as shown in
In this example, the second clock signal, clk2, may be skewed (i.e., delayed) relative to the first clock signal, clk1, by a certain amount, e.g., skew [0-2]-skew [0-1]. The Advantest T5585 can provide, for example, a periodic signal of pulses on several channels, where each pulse has a minimum pulse width of about 1 ns at a minimum repetition interval of about 4 ns, up to a maximum pulse width of 0.5 ms at a maximum repetition interval of about 4 ms. The pulse forming circuit (described in detail below) starts a word line signal, WL, based on the rising edge of the first clock signal, clk1, and stops the word line signal, WL, based on the rising edge of the second clock signal, clk2. The skew delay between the first clock signal, clk1, and the second clock signal, clk2, from the Advantest T5585 may range from approximately 0.1 ns in the 4 ns repetition interval range to approximately 0.9 ms in the 4 ms repetition interval range. Thus, it can be seen that while the tester may not be able to provide a pulse less than about 1 ns in duration, it can provide two skewed clock pulses that may be applied to the pulse forming circuit to form a pulse in the word line signal, WL, ranging approximately from less than 1 ns to 0.9 ms in duration.
In addition, as described in more detail below, a bypass signal, BYPASS, may be provided to the pulse forming circuit that, when applied, may be provides a word line signal, WL, that is a time replica of the third clock signal, clk3, overriding the pulse determined by the skew between the first clock signal, clk1, and the second clock signal, clk2. The third clock signal, clk3, may be a pulse of any duration, including up to DC. The word line signal, WL, formed by either the relative skew of first clock signal, clk1 and the second clock signal, clk2, or by the third clock signal, clk3, may be the applied to the gate G of the transistor 202 of the particular MTJ bitcell 200 under test.
When properly triggered, the latch 401 becomes transparent and passes the first clock signal, clk1, to a set-reset latch SR 403. When properly triggered, the latch 402 becomes transparent and passes the second clock signal, clk2, first to an OR gate 404 and then to the SR latch 403. The latch 401 and the latch 402 are triggered to become transparent when respectively receiving the first clock enable signal, clk1_enable, and second clock enable signal, clk2_enable. When the latch 401 receives the first clock enable signal, clk1_enable, becoming transparent to the first clock signal, clk1, the first clock signal, clk1, is applied to an input SET of the SR latch 403. A rising edge of the first clock signal, clk1, sets the SR latch 403, making it transparent, and the first clock signal, clk1, passes through. Similarly, when the SR latch 402 receives the second clock enable signal, clk2_enable, becoming transparent to the second clock signal, clk2, the second clock signal, clk2, is applied, via the OR gate 404, to a reset input RST of the SR latch 403 to reset the SR latch 403. A high second clock signal, clk2, such as the rising edge of the second clock signal, clk2, resets the SR latch 403 and makes it opaque, ending the signal passing through the SR latch 403. Thus, a pulse signal, pulse, has a pulse width determined by the delay between the leading edges of the first clock signal, clk1, and the second clock signal, clk2. The delay may be variable, and of much shorter duration than a pulse length of either of the first and second clock signals, clk1 and clk2.
Additionally, the latch 401, the latch 402 and the OR gate 404 have inputs for a Global Reset signal. When the Global Reset signal goes high (i.e., a logical “1”), both latches 401 and 402 provide an output that goes low. The OR gate 404 receives the high Global Reset signal and accordingly resets the SR latch 403 to output a low signal, i.e., no pulse. The Global Reset signal performs the function of starting each test measurement sequence with the pulse signal, pulse, low, i.e., closing the gate G so that the transistor 202 is an open circuit, and the state of the MTJ bitcell 200 can be neither written to nor read from.
A multiplexer, mux 405 is configured to receive the output signal, pulse, from the SR latch 403, the third clock signal, clk3, and the bypass signal, BYPASS. The output signal from the mux 405 depends on the level of the bypass signal, BYPASS. When the bypass signal, BYPASS, is set high, i.e., to “1”, the mux 405 outputs the third clock signal, clk3. When the bypass signal, BYPASS, is high, the mux 405 outputs a pulse word line, WL, of pulse width determined by the third clock signal, clk3. Thus, the mux 405 enables the pulse width forming portion of the circuit 400 to be substantially bypassed if, for example, there is a malfunction in any of the corresponding pulse forming circuitry (e.g., the latches 401, 402, 403 and the OR gate 404). Pulse forming and testing functionality is retained to the extent that a word line signal, WL, based on the third clock signal, clk3, is available. Conversely, when the bypass signal, BYPASS, is set low (“0”), the mux 405 outputs the a word line signal WL, based on the pulse output from the SR latch 403. In one embodiment, local word lines and global word lines can be used. In this case, whichever signal is output from the mux 405, as determined by the bypass signal, BYPASS, may be ANDed with a global word line signal (not shown) to provide a local word line signal (not shown) to the gate G of the transistor 201 of the MTJ bitcell 200 being tested.
Prior to a time a, the first clock signal, clk1, first clock enable signal, clk1_enable, second clock signal, clk2, and second clock enable signal, clk2_enable are initialized to “off” or “low” states (e.g., logical “0”), which may be, for example, zero volts, and a variable bit line voltage VBL and a variable source line voltage VSL are also low, so that there is substantially no voltage difference applied across an MTJ bitcell 200. At the time a, there is no signal emitted from the latch 403. At the time a, the bit line voltage VBL level is set to a first value (e.g., 2.5 volts), while the source line voltage VSL remains low. The voltage difference VBL-VSL is large enough to write a state 0 polarization to the MTJ 201. However, since there is no word line signal, WL, to open the gate G of the transistor 202 to permit current flow, no polarization state is yet written.
Between time a and time b, the first clock enable signal, clk1_enable, and the second clock enable signal, clk2_enable, go high, so that the first clock signal, clk1 and and the second clock signal, clk2, pulses can be sampled. At time b, the first clock signal, clk1, goes high, taking the pulse line signal, pulse, high. Since the word line signal, WL, is the image of the pulse line signal, pulse, the word line signal, WL, goes high to open the gate G of the transistor 202. Because the relative voltage VBL-VSL is configured to write a state 0, the word line signal, WL, is here referred to as a w0 pulse during this time interval. At time c, the second clock signal, clk2, goes high, taking the pulse signal, pulse, low, and simultaneously ending the w0 pulse.
Between time c and time d, the bit line voltage, VBL, is reduced to a level so that the voltage difference VBL-VSL is small enough to read the resistance of the MTJ bitcell 200 but not to change the state of the MTJ 201. At time d, the first clock signal, clk1, goes high, taking the pulse signal, pulse, output from the latch 402 high simultaneously with the word line signal, WL, to open the gate G again. Because the voltage across the MTJ bitcell 200 is the smaller read mode value, the word line signal, WL, is referred to here as an r0 pulse during the time interval d-e. At time e, the second clock signal, clk2, goes high, taking the pulse signal output from the latch 402 low, ending the word line signal, WL, r0 pulse.
During the time interval between c and f, the bit line voltage signal, VBL, goes low, and the source line voltage, VSL, goes high, substantially reversing the write mode voltage across the MTJ bitcell 200 that occurred during the interval b-c. This corresponds to the voltage applied to write a state 1 polarization to the MTJ 201. As described above, the first clock signal, clk1, goes high at time f and the second clock signal, clk2, goes high at time g. During this time interval, the pulse signal, pulse, output from the latch 403 is provided and the word line signal, WL, opens the gate of the transistor 202, writing the state 1 polarization to the MTJ 201. This write signal is referred to here as a w1 pulse, because a state 1 polarization is being written.
During the time interval g-h, the bit line voltage VBL, is changed to the same value selected in time interval c-d, and the source line voltage, VSL, is taken low, again as in interval c-d, with the result that a read mode level of voltage is applied to the MTJ bitcell 200 when the gate G of the transistor 202 is open. During time interval h-i, the first and second clock signals, clk1, and clk2, again operate to provide the signal rising edges to form the word line signal, WL, as a pulse of the time duration h-i. Because the MTJ 201 has a state 1, the pulse formed in the word line signal, WL, to read the MTJ 201 resistance at this time is referred to as an r1 pulse.
The output pulse from the SR latch 403 is then input to the mux 405 which, because the bypass signal, BYPASS, is low, outputs a word line signal, WL, of length corresponding to the pulse output from the SR latch 403. The word line signal, WL, is then output to the MTJ bitcell 200 of the STT-MRAM memory (block 826). The word line signal, WL, for example, is applied to the gate G of the transistor 202 of the MTJ bitcell 200. In one embodiment, a local word line signal is provided as an output of an AND gate by ANDing the word line signal, WL, with a global word line signal.
If the bypass signal, BYPASS, is high (in decision block 820), the mux 405 admits the third clock signal, clk3, (block 831), and then outputs a word line signal, WL, with a pulse length corresponding to the third clock signal, clk3 (block 832). Again, a local word line signal can be created by ANDing the word line signal, WL, with a global word line signal. The local word line signal can be output to the MTJ bitcell 200 of the STT-MRAM memory, as described above.
Although specific circuitry has been set forth, it will be appreciated by those skilled in the art that not all of the disclosed circuitry is required to practice the disclosure. Moreover, certain well known circuits have not been described, to maintain focus on the disclosure. Similarly, although the description refers to logical “0” or “low” and logical “1” or “high” in certain locations, one skilled in the art appreciates that the logical values can be switched, with the remainder of the circuit adjusted accordingly, without affecting operation of the present disclosure.
Data recorded on the storage medium 904 may specify logic circuit configurations, pattern data for photolithography masks, or mask pattern data for serial write tools such as electron beam lithography. The data may further include logic verification data such as timing diagrams or net circuits associated with logic simulations. Providing data on the storage medium 904 facilitates the design of the circuit design 910 by decreasing the number of processes for designing semiconductor ICs.
Although the present disclosure and its advantages have been described in detail, it should be understood that various changes, substitutions and alterations can be made herein without departing from the spirit and scope of the disclosure as defined by the appended claims. For example, formation of the word line pulse has been described in terms of triggering the start and end of the pulse based on the relative skew of the rising edges of the first and second clock signal, clk1 and clk2. Alternatively, formation of the word line signal, WL, may be determined from the relative skew of the falling edges of the the first and second clock signal, clk1 and clk2 or any combination of rising and falling edges. Moreover, the scope of the present application is not intended to be limited to the particular embodiments of the process, machine, manufacture, composition of matter, means, methods and steps described in the specification. As one of ordinary skill in the art will readily appreciate from the present disclosure, processes, machines, manufacture, compositions of matter, means, methods, or steps, presently existing or later to be developed that perform substantially the same function or achieve substantially the same result as the corresponding embodiments described herein may be utilized according to the present disclosure. Accordingly, the appended claims are intended to include within their scope such processes, machines, manufacture, compositions of matter, means, methods, or steps.
| Number | Name | Date | Kind |
|---|---|---|---|
| 4994687 | Fujii et al. | Feb 1991 | A |
| 5422585 | Fan Chiangi et al. | Jun 1995 | A |
| 5959485 | Roohparvar | Sep 1999 | A |
| 6972998 | Gibson et al. | Dec 2005 | B1 |
| 7224639 | Lee | May 2007 | B2 |
| 7349270 | Sawhney | Mar 2008 | B2 |
| 20060097755 | Yamanaka | May 2006 | A1 |
| Number | Date | Country | |
|---|---|---|---|
| 20110051502 A1 | Mar 2011 | US |
